This invention relates to routing data on a computer network.
A typical computer network is collection of interconnected computers. Data, such as text, images, sounds and videos, are typically communicated over the network from a source computer to a destination computer in pieces referred to as “packets.” In most networks, packets usually are not sent directly from the source to the destination, but rather are routed by intermediate nodes known as “routers.”
Generally, there are two types of packets that a router may receive. First, a general data packet contains data for communication between two computers on the network. The router handles this type of data packet by forwarding the packet to the appropriate destination computer or next router as necessary.
Second, control information packets carry control information and are used by the router itself. Control information includes router-to-router communications and may also include administrative network commands. Changes of router configuration, such as management of communication ports and changing the internet protocol (IP) addresses of the router communication ports, as well as changes of network topology, are examples of control information. Good network performance requires fast packet forwarding, but rapid exchange of control information is less critical to network performance.
To improve the speed of packet forwarding, many routers employ network processors, which are specialized hardware devices designed to perform certain networking operations quickly. A network processor operates separately from the router's general-purpose processor and usually isolates the operating system on the general-purpose processor from the forwarding operations. A router's network processor typically has more limited, dedicated resources than the router's general-purpose processor.